8.4.10 2-Categorical Epimorphisms

    In this section we characterise (for now, some of) the $2$-categorical epimorphisms in $\boldsymbol {\mathsf{Rel}}$, following Chapter 13: Types of Morphisms in Bicategories, Section 13.2.

    Let $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ be a relation.

    1. 1.

      Corepresentably Faithful Morphisms in $\boldsymbol {\mathsf{Rel}}$. Every morphism of $\boldsymbol {\mathsf{Rel}}$ is a corepresentably faithful morphism.

    2. 2.

      Corepresentably Full Morphisms in $\boldsymbol {\mathsf{Rel}}$. The following conditions are equivalent:

      1. (a)

        The morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is a corepresentably full morphism.

      2. (b)

        For each pair of relations $S,T\colon X\mathrel {\rightrightarrows \kern -9.5pt\mathrlap {|}\kern 6pt}A$, the following condition is satisfied:

        • (★)
        • If $S\mathbin {\diamond }R\subset T\mathbin {\diamond }R$, then $S\subset T$.
      3. (c)

        The functor

        \[ R^{-1}\colon \webleft (\mathcal{P}\webleft (B\webright ),\subset \webright )\to \webleft (\mathcal{P}\webleft (A\webright ),\subset \webright ) \]

        is full.

      4. (d)

        For each $U,V\in \mathcal{P}\webleft (B\webright )$, if $R^{-1}\webleft (U\webright )\subset R^{-1}\webleft (V\webright )$, then $U\subset V$.

      5. (e)

        The functor

        \[ R_{-1}\colon \webleft (\mathcal{P}\webleft (B\webright ),\subset \webright )\to \webleft (\mathcal{P}\webleft (A\webright ),\subset \webright ) \]

        is full.

      6. (f)

        For each $U,V\in \mathcal{P}\webleft (B\webright )$, if $R_{-1}\webleft (U\webright )\subset R_{-1}\webleft (V\webright )$, then $U\subset V$.

  • 3.

    Corepresentably Fully Faithful Morphisms in $\boldsymbol {\mathsf{Rel}}$. Every corepresentably full morphism of $\boldsymbol {\mathsf{Rel}}$ is a corepresentably fully faithful morphism.

  • Item 1: Corepresentably Faithful Morphisms in $\boldsymbol {\mathsf{Rel}}$
    The relation $R$ is a corepresentably faithful morphism in $\boldsymbol {\mathsf{Rel}}$ iff , for each $X\in \operatorname {\mathrm{Obj}}\webleft (\boldsymbol {\mathsf{Rel}}\webright )$, the functor

    \[ R^{*}\colon \mathbf{Rel}\webleft (B,X\webright )\to \mathbf{Rel}\webleft (A,X\webright ) \]

    is faithful, i.e. iff the morphism

    \[ R^{*}_{S,T}\colon \operatorname {\mathrm{Hom}}_{\mathbf{Rel}\webleft (B,X\webright )}\webleft (S,T\webright )\to \operatorname {\mathrm{Hom}}_{\mathbf{Rel}\webleft (A,X\webright )}\webleft (S\mathbin {\diamond }R,T\mathbin {\diamond }R\webright ) \]

    is injective for each $S,T\in \operatorname {\mathrm{Obj}}\webleft (\mathbf{Rel}\webleft (B,X\webright )\webright )$. However, $\operatorname {\mathrm{Hom}}_{\mathbf{Rel}\webleft (B,X\webright )}\webleft (S,T\webright )$ is either empty or a singleton, in either case of which the map $R^{*}_{S,T}$ is necessarily injective.

    Item 2: Corepresentably Full Morphisms in $\boldsymbol {\mathsf{Rel}}$
    We claim Item 2a, Item 2b, Item 2c, Item 2d, Item 2e, and Item 2f are indeed equivalent:

    • Item 2a$\iff $Item 2b: This is simply a matter of unwinding definitions: The relation $R$ is a corepresentably full morphism in $\boldsymbol {\mathsf{Rel}}$ iff , for each $X\in \operatorname {\mathrm{Obj}}\webleft (\boldsymbol {\mathsf{Rel}}\webright )$, the functor

      \[ R^{*}\colon \mathbf{Rel}\webleft (B,X\webright )\to \mathbf{Rel}\webleft (A,X\webright ) \]

      is full, i.e. iff the morphism

      \[ R^{*}_{S,T}\colon \operatorname {\mathrm{Hom}}_{\mathbf{Rel}\webleft (B,X\webright )}\webleft (S,T\webright )\to \operatorname {\mathrm{Hom}}_{\mathbf{Rel}\webleft (A,X\webright )}\webleft (S\mathbin {\diamond }R,T\mathbin {\diamond }R\webright ) \]

      is surjective for each $S,T\in \operatorname {\mathrm{Obj}}\webleft (\mathbf{Rel}\webleft (B,X\webright )\webright )$, i.e. iff , whenever $S\mathbin {\diamond }R\subset T\mathbin {\diamond }R$, we also have $S\subset T$.

    • Item 2c$\iff $Item 2d: This is also simply a matter of unwinding definitions: The functor

      \[ R^{-1}\colon \webleft (\mathcal{P}\webleft (B\webright ),\subset \webright )\to \webleft (\mathcal{P}\webleft (A\webright ),\subset \webright ) \]

      is full iff , for each $U,V\in \mathcal{P}\webleft (A\webright )$, the morphism

      \[ R^{-1}_{U,V}\colon \operatorname {\mathrm{Hom}}_{\mathcal{P}\webleft (B\webright )}\webleft (U,V\webright )\to \operatorname {\mathrm{Hom}}_{\mathcal{P}\webleft (A\webright )}\webleft (R^{-1}\webleft (U\webright ),R^{-1}\webleft (V\webright )\webright ) \]

      is surjective, i.e. iff whenever $R^{-1}\webleft (U\webright )\subset R^{-1}\webleft (V\webright )$, we also necessarily have $U\subset V$.

    • Item 2e$\iff $Item 2f: This is once again simply a matter of unwinding definitions, and proceeds exactly in the same way as in the proof of the equivalence between Item 2c and Item 2d given above.

    • Item 2d$\implies $Item 2f: Suppose that the following condition is true:

      • (★)
      • For each $U,V\in \mathcal{P}\webleft (B\webright )$, if $R^{-1}\webleft (U\webright )\subset R^{-1}\webleft (V\webright )$, then $U\subset V$.

      We need to show that the condition

      • (★)
      • For each $U,V\in \mathcal{P}\webleft (B\webright )$, if $R_{-1}\webleft (U\webright )\subset R_{-1}\webleft (V\webright )$, then $U\subset V$.

      is also true. We proceed step by step:

    • Item 2f$\implies $Item 2d: Suppose that the following condition is true:

      • (★)
      • For each $U,V\in \mathcal{P}\webleft (B\webright )$, if $R_{-1}\webleft (U\webright )\subset R_{-1}\webleft (V\webright )$, then $U\subset V$.

      We need to show that the condition

      • (★)
      • For each $U,V\in \mathcal{P}\webleft (B\webright )$, if $R^{-1}\webleft (U\webright )\subset R^{-1}\webleft (V\webright )$, then $U\subset V$.

      is also true. We proceed step by step:

    • Item 2b$\implies $Item 2d: Consider the diagram

      and suppose that $S\mathbin {\diamond }R\subset T\mathbin {\diamond }R$. Note that, by assumption, given a diagram of the form
      if $R^{-1}\webleft (U\webright )=R\mathbin {\diamond }U\subset R\mathbin {\diamond }V=R^{-1}\webleft (V\webright )$, then $U\subset V$. In particular, for each $x\in X$, we may consider the diagram
      for which we have $\webleft [x\webright ]\mathbin {\diamond }S\mathbin {\diamond }R\subset \webleft [x\webright ]\mathbin {\diamond }T\mathbin {\diamond }R$, implying that we have

      \[ S^{-1}\webleft (x\webright )=\webleft [x\webright ]\mathbin {\diamond }S\subset \webleft [x\webright ]\mathbin {\diamond }T=T^{-1}\webleft (x\webright ) \]

      for each $x\in X$, implying $S\subset T$.

    • Item 2d$\implies $Item 2b: Let $U,V\in \mathcal{P}\webleft (B\webright )$ and consider the diagram

      By Remark 8.5.1.1.2, we have

      \begin{align*} R^{-1}\webleft (U\webright ) & = U\mathbin {\diamond }R,\\ R^{-1}\webleft (V\webright ) & = V\mathbin {\diamond }R. \end{align*}

      Now, if $R^{-1}\webleft (U\webright )\subset R^{-1}\webleft (V\webright )$, i.e. $U\mathbin {\diamond }R\subset V\mathbin {\diamond }R$, then $U\subset V$ by assumption.

    Item 3: Corepresentably Fully Faithful Morphisms in $\boldsymbol {\mathsf{Rel}}$
    This follows from Item 1 and Item 2.


Noticed something off, or have any comments? Feel free to reach out!


You can also use the contact form below: